Filtering the cardiopulmonary resuscitation (CPR) artifact has been a major approach to minimizing interruptions to CPR for rhythm analysis. However, the effects of these filters on interruptions to CPR have not been evaluated. This study presents the first methodology for directly quantifying the effects of filtering on the uninterrupted CPR time.A total of 241 shockable and 634 nonshockable out-of-hospital cardiac arrest records (median duration, 150 seconds) from 248 patients were analyzed. Filtering and rhythm analysis were commenced after 1 minute of CPR, and the end point for CPR was established at the time of the first shock diagnosis. Kaplan-Meier curves were used to compute the probability of interrupting CPR as a function of time. The probabilities of delivering 2 minutes of uninterrupted CPR for the shockable and nonshockable rhythms were compared with the 2-minute cycles of uninterrupted CPR recommended by the guidelines.For the nonshockable rhythms, the probabilities of delivering at least 2 and 3 minutes of uninterrupted CPR were 58% (95% confidence interval, 54%-62%) and 48% (44%-52%), respectively. These are the probabilities of reducing and substantially reducing the frequency of CPR interruptions for rhythm analysis. For the shockable rhythms, the probability of avoiding unnecessary CPR prolongation beyond 2 minutes was 100% (99%-100%).Filtering reduces the frequency of CPR interruptions for rhythm analysis in less than 60% of nonshockable rhythms. New strategies to increase the probability of prolonging CPR for nonshockable rhythms should be defined and evaluated using the methodology proposed in this study.